In my experience as a devops/full-stack dev at a boutique SaaS company, I realized that logs of web applications are the first point of detecting and debugging problems, or just being updated (say in case someone is DDoS-ing your app) but it was also the most difficult thing for the rest of my team to do (say, non-unix-y people, operations head, etc.). And I've been sitting on this idea for a while and I'm curious what you think.
Pitch- Suppose you've an app (Django, RoR, Flask, Apache, etc.) running (in a container/VM/server), you can access each registered apps logs over the web through my service on your servers IP address. You can additionally apply quick filters, alerts (premium), etc. In a later version, I'd ideally want to add framework specific alerts, like MongoDB is running without password, or Django is hitting some bad URL endpoint, etc.
Bonus- it's a lot like Kibana, Filebeat, ElasticSearch, except, with my service you don't have to be a pro and don't have to learn this whole mess. You just write one line (something like: python logtoss -b) and you'll be all set! Is this something that you'd find useful? Would you pay for using an enterprise solution (i.e. with support, and alerts, and the latest updates)?
My main concern is that I am running your agent on my server:
Maybe open sourcing it and having a hosted solution with support could be the way to go (mitigating the ealier points).
My be I was not clear-
You don't open a port to a different server, but you listen on a port on your own server, but that is protected by your own credentials that is available to you alone.
You're right regarding the privacy concerns.
I'm actually considering open-sourcing it in the beginning.
I meant opening a port is somewhat of a no-no, and you will need to market the product with that in mind.
Why not open sourcing it forever and having customer support + premium features paid? (gitlab, most of the DBs out there, Ghost blogs...)